DLTJob.cs 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359
  1. using System;
  2. using System.Collections.Generic;
  3. using System.Linq;
  4. using System.Text;
  5. using HtmlAgilityPack;
  6. using Quartz;
  7. using SCC.Common;
  8. using SCC.Crawler.Tools;
  9. using SCC.Interface;
  10. using SCC.Models;
  11. namespace SCC.Crawler.QG
  12. {
  13. /// <summary>
  14. /// 全国彩大乐透爬取类
  15. /// 创建人·彭怀畅
  16. /// 时间·13点25分
  17. /// </summary>
  18. public class DLTJob : IJob
  19. {
  20. /// <summary>
  21. /// 构造函数
  22. /// </summary>
  23. public DLTJob()
  24. {
  25. log = new LogHelper();
  26. services = IOC.Resolve<IDTOpenCode>();
  27. email = IOC.Resolve<IEmail>();
  28. }
  29. /// <summary>
  30. /// 作业执行入口
  31. /// </summary>
  32. /// <param name="context">作业执行上下文</param>
  33. public void Execute(IJobExecutionContext context)
  34. {
  35. Config = CommonHelper.GetConfigFromDataMap(context.JobDetail.JobDataMap);
  36. //预设节假日不开奖
  37. if (Config.SkipDate.Contains(CommonHelper.SCCSysDateTime.ToString("yyyyMMdd"))) return;
  38. LatestItem = context.JobDetail.JobDataMap["LatestItem"] as OpenCode7DTModel;
  39. try
  40. {
  41. //服务启动时配置初始数据
  42. if (LatestItem == null)
  43. {
  44. LatestItem = services.GetOpenCode7DTLastItem(currentLottery);
  45. //如果数据库没有数据和期号就自动生成期号
  46. if (LatestItem == null)
  47. LatestItem = new OpenCode7DTModel
  48. {
  49. Term = CommonHelper.GenerateQiHaoYYYYQQQ(0),
  50. OpenTime = new DateTime(CommonHelper.SCCSysDateTime.Year, 1, 1)
  51. };
  52. }
  53. //程序时间第二天,程序根据配置检查是否昨天有开奖
  54. isGetData = false;
  55. if (CommonHelper.CheckDTIsNeedGetData(Config)) DoMainUrl();
  56. if (!LatestItem.Term.ToString().StartsWith(CommonHelper.SCCSysDateTime.ToString("yy")))
  57. LatestItem = new OpenCode7DTModel
  58. {
  59. Term = CommonHelper.GenerateQiHaoYYYYQQQ(0),
  60. OpenTime = new DateTime(CommonHelper.SCCSysDateTime.Year, 1, 1)
  61. };
  62. //当今日开奖并且当前时间是晚上8点过后开始抓取
  63. if (CommonHelper.CheckTodayIsOpenDay(Config) && CommonHelper.SCCSysDateTime.Hour > 12) DoMainUrl();
  64. CheckGetKaiJiHao();
  65. }
  66. catch (Exception ex)
  67. {
  68. log.Error(GetType(), string.Format("【{0}】抓取时发生错误,错误信息【{1}】", Config.Area + currentLottery, ex.Message));
  69. }
  70. //保存最新期号
  71. context.JobDetail.JobDataMap["LatestItem"] = LatestItem;
  72. }
  73. /// <summary>
  74. /// 更新开机号
  75. /// </summary>
  76. private void CheckGetKaiJiHao()
  77. {
  78. try
  79. {
  80. List<KaiJiangHao> kai = GetKaiJiHao();
  81. foreach (var item in kai)
  82. {
  83. var isSucc = services.UpdateSSQDetailByTerm(currentLottery, item.QiHao, item.Kaijianghao);
  84. if (isSucc)
  85. {
  86. log.Info(GetType(), $"更新{Config.LotteryName}第{ item.QiHao}期开机号成功!");
  87. }
  88. else
  89. {
  90. log.Error(GetType(), $"更新{Config.LotteryName}第{ item.QiHao}期开机号失败!");
  91. }
  92. }
  93. }
  94. catch (Exception e)
  95. {
  96. log.Error(GetType(), e);
  97. }
  98. }
  99. /// <summary>
  100. /// 获取开机号
  101. /// </summary>
  102. /// <returns></returns>
  103. private List<KaiJiangHao> GetKaiJiHao()
  104. {
  105. List<KaiJiangHao> result = new List<KaiJiangHao>();
  106. var url = "https://www.8200.cn/kjh/ssq/kjih.htm?size=30";
  107. var htmlResource = NetHelper.GetUrlResponse(url, Encoding.GetEncoding("utf-8"));
  108. if (htmlResource == null) return null;
  109. if (!string.IsNullOrEmpty(htmlResource))
  110. {
  111. var doc = new HtmlDocument();
  112. doc.LoadHtml(htmlResource);
  113. var table = doc.DocumentNode.SelectSingleNode("//table");
  114. if (table == null) return null;
  115. var trs = table.ChildNodes.Where(node => node.Name == "tr").ToList();
  116. KaiJiangHao model = null;
  117. for (var i = 0; i <= 15; i++)
  118. {
  119. var tds = trs[i].ChildNodes.Where(S => S.Name.ToLower() == "td").ToList();
  120. model = new KaiJiangHao();
  121. model.QiHao = Convert.ToInt32(tds[0].InnerText.Trim());
  122. if (tds[2].InnerText.Trim() == "--")
  123. {
  124. model.Kaijianghao = "";
  125. }
  126. else
  127. {
  128. string source = tds[2].InnerText.Replace(" + ", ",").Replace(" ", ",").Replace(",,", ",").Replace(",,", ",").Trim();
  129. source = source.IndexOf(",") <= 0 ? source.Substring(1, source.Length - 1) : source;
  130. source = source.LastIndexOf(",") >= 0 ? source.Substring(0, 20) : source;
  131. model.Kaijianghao = source;
  132. }
  133. result.Add(model);
  134. }
  135. }
  136. return result;
  137. }
  138. /// <summary>
  139. /// 通过主站点爬取开奖数据
  140. /// (福建体彩网)
  141. /// </summary>
  142. private void DoMainUrl()
  143. {
  144. if (!string.IsNullOrEmpty(Config.MainUrl))
  145. {
  146. var openList = GetOpenListFromMainUrl(Config.MainUrl);
  147. if (openList.Count == 0) return; //无抓取数据
  148. //抓取到的最新期数
  149. var newestQiHao = Convert.ToInt32(openList.First().Term.ToString());
  150. //数据库里面最新期数
  151. var startQiNum = Convert.ToInt32(LatestItem.Term.ToString());
  152. if (startQiNum > newestQiHao) return; //无最新数据
  153. //处理最新开奖数据
  154. var getQiHao = string.Empty;
  155. OpenCode7DTModel matchItem = null;
  156. for (var i = startQiNum; i <= newestQiHao; i++)
  157. {
  158. getQiHao = i.ToString();
  159. matchItem = openList.FirstOrDefault(r => r.Term.ToString() == getQiHao);
  160. if (matchItem != null && services.AddDTOpen7Code(currentLottery, matchItem))
  161. {
  162. //Do Success Log
  163. log.Info(GetType(), CommonHelper.GetJobMainLogInfo(Config, getQiHao));
  164. LatestItem = matchItem;
  165. isGetData = true;
  166. }
  167. }
  168. }
  169. }
  170. /// <summary>
  171. /// 根据主站获取数据列表
  172. /// </summary>
  173. /// <param name="mainUrl"></param>
  174. /// <returns></returns>
  175. private List<OpenCode7DTModel> GetOpenListFromMainUrl(string mainUrl)
  176. {
  177. var result = new List<OpenCode7DTModel>();
  178. try
  179. {
  180. var url = new Uri(mainUrl);
  181. var htmlResource = NetHelper.GetUrlResponse(mainUrl, Encoding.GetEncoding("utf-8"));
  182. if (htmlResource == null) return result;
  183. var doc = new HtmlDocument();
  184. doc.LoadHtml(htmlResource);
  185. var table = doc.DocumentNode.SelectSingleNode("//table");
  186. if (table == null) return result;
  187. var trs = table.ChildNodes.Where(node => node.Name == "tr").ToList();
  188. OpenCode7DTModel model = null;
  189. HtmlNode nodeA = null;
  190. var optimizeUrl = string.Empty;
  191. for (var i = 0; i < trs.Count; i++) //第一二行为表头
  192. {
  193. var tds = trs[i].ChildNodes.Where(node => node.Name == "td").ToList();
  194. if (tds.Count < 9) continue;
  195. model = new OpenCode7DTModel();
  196. var Term = tds[0].InnerText.Trim();
  197. if (Term.Length <= 7)
  198. {
  199. model.Term = Convert.ToInt64("20" + Term);
  200. }
  201. else
  202. {
  203. model.Term = Convert.ToInt64(Term);
  204. }
  205. optimizeUrl = model.Term.ToString();
  206. model.OpenTime = Convert.ToDateTime(tds[1].InnerText.Substring(0, 5));
  207. if (tds[2].ChildNodes.Count == 0) continue;
  208. var b = tds[2].ChildNodes.Where(n => n.Name.ToLower() == "b").ToList();
  209. var span = b[0].ChildNodes.Where(n => n.Name.ToLower() == "span").ToList();
  210. model.OpenCode1 = Convert.ToInt32(span[0].InnerText.Replace(" ", "").Substring(0, 2).Trim());
  211. model.OpenCode2 = Convert.ToInt32(span[0].InnerText.Replace(" ", "").Substring(2, 2).Trim());
  212. model.OpenCode3 = Convert.ToInt32(span[0].InnerText.Replace(" ", "").Substring(4, 2).Trim());
  213. model.OpenCode4 = Convert.ToInt32(span[0].InnerText.Replace(" ", "").Substring(6, 2).Trim());
  214. model.OpenCode5 = Convert.ToInt32(span[0].InnerText.Replace(" ", "").Substring(8, 2).Trim());
  215. model.OpenCode6 = Convert.ToInt32(span[1].InnerText.Replace(" ", "").Substring(0, 2).Trim());
  216. model.OpenCode7 = Convert.ToInt32(span[1].InnerText.Replace(" ", "").Substring(2, 2).Trim());
  217. var details = GetKaijiangDetails(optimizeUrl);
  218. model.Spare = details;
  219. result.Add(model);
  220. }
  221. var checkDataHelper = new CheckDataHelper();
  222. var dbdata = services.GetListS<OpenCode7DTModel>(currentLottery)
  223. .ToDictionary(w => w.Term.ToString(), w => w.GetCodeStr());
  224. checkDataHelper.CheckData(dbdata, result.ToDictionary(w => w.Term.ToString(), w => w.GetCodeStr()),
  225. Config.Area, currentLottery);
  226. result = result.OrderByDescending(S => S.Term).ToList();
  227. }
  228. catch (Exception ex)
  229. {
  230. log.Error(GetType(),
  231. string.Format("【{0}】通过主站抓取开奖列表时发生错误,错误信息【{1}】", Config.Area + currentLottery, ex.Message));
  232. }
  233. return result;
  234. }
  235. /// <summary>
  236. /// 开奖详情
  237. /// </summary>
  238. /// <param name="optimizeUrl"></param>
  239. /// <returns></returns>
  240. private string GetKaijiangDetails(string optimizeUrl)
  241. {
  242. var url = "https://www.8200.cn/kjh/dlt/" + optimizeUrl + ".htm";
  243. var htmlResource = NetHelper.GetUrlResponse(url, Encoding.GetEncoding("utf-8"));
  244. var doc = new HtmlDocument();
  245. doc.LoadHtml(htmlResource);
  246. var div = doc.DocumentNode.SelectSingleNode("//div[@class='text-16']");
  247. if (div == null) return null;
  248. //爬去奖金
  249. var jiangjin = div.ChildNodes.Where(node => node.Name == "p").ToList();
  250. //爬去奖项
  251. //var tbody = div.ChildNodes.Where(node => node.Name == "tbody").ToList();
  252. var table = doc.DocumentNode.SelectSingleNode("//table");
  253. var trs = table.ChildNodes.Where(node => node.Name == "tr").ToList();
  254. var gdje = jiangjin[3].InnerText.Replace(" 亿元", "").Replace("奖池滚存:", "").Replace("--", "0").Replace(",", "")
  255. .Trim();
  256. var trje = jiangjin[2].InnerText.Replace(" 亿元", "").Replace("本期销量:", "").Replace("--", "0").Replace(",", "")
  257. .Trim();
  258. var entity = new KaijiangDetailsEntity
  259. {
  260. Gdje = gdje == "0" ? "0" : (double.Parse(gdje) * 100000000).ToString(),
  261. Trje = trje == "0" ? "0" : (double.Parse(trje) * 100000000).ToString()
  262. };
  263. //TODO
  264. //组装详情
  265. var list = new List<Kaijiangitem>();
  266. for (var i = 0; i < trs.Count; i++)
  267. {
  268. var tds = trs[i].ChildNodes.Where(node => node.Name == "td").ToList();
  269. var kaijiangitem = new Kaijiangitem();
  270. var TotalMoney = tds[1].InnerText.Replace("元", "").Replace("--", "0").Replace(",", "").Trim();
  271. kaijiangitem.Name = tds[0].InnerText.Trim();
  272. kaijiangitem.TotalMoney = TotalMoney == "0" ? "0" : double.Parse(TotalMoney).ToString();
  273. kaijiangitem.Total = tds[2].InnerText.Replace(" 注", "").Replace("--", "0").Trim();
  274. list.Add(kaijiangitem);
  275. }
  276. entity.KaiJiangItems = list;
  277. return entity.TryToJson();
  278. }
  279. #region Attribute
  280. /// <summary>
  281. /// 配置信息
  282. /// </summary>
  283. private SCCConfig Config;
  284. /// <summary>
  285. /// 当天抓取的最新一期开奖记录
  286. /// </summary>
  287. private OpenCode7DTModel LatestItem;
  288. /// <summary>
  289. /// 当天抓取失败列表
  290. /// </summary>
  291. private List<string> FailedQiHaoList = null;
  292. /// <summary>
  293. /// 日志对象
  294. /// </summary>
  295. private readonly LogHelper log;
  296. /// <summary>
  297. /// 数据服务
  298. /// </summary>
  299. private readonly IDTOpenCode services;
  300. /// <summary>
  301. /// 当前彩种
  302. /// </summary>
  303. private SCCLottery currentLottery => SCCLottery.DLT;
  304. /// <summary>
  305. /// 邮件接口
  306. /// </summary>
  307. private IEmail email;
  308. /// <summary>
  309. /// 是否本次运行抓取到开奖数据
  310. /// </summary>
  311. private bool isGetData;
  312. #endregion
  313. }
  314. }